CN100461454C - Semiconductor device and method for forming the same - Google Patents

Abstract

The invention provides a semiconductor element and a method for forming the same. The gate comprises a gate electrode and a gate dielectric layer, and the gate dielectric layer is located under the gate electrode. The spacer walls form sidewalls of the gate electrode and the gate dielectric layer. The buffer layer is located on a semiconductor substrate, the buffer layer has a second part located under the gate dielectric layer and the spacer, and has a first part adjacent to the spacer, wherein the upper surface of the buffer layer located at the first part is recessed from the upper surface of the buffer layer located at the second part. The source/drain regions are substantially aligned with the spacers. The lattice constant of the buffer layer is larger than that of the underlying substrate. The semiconductor element further comprises a semiconductor covering layer positioned between the buffer layer and the grid dielectric layer, wherein the lattice constant of the semiconductor covering layer is smaller than that of the buffer layer.

Description

Semiconductor element and method for forming semiconductor element

technical field

The present invention relates to a semiconductor, in particular to a complementary metal-oxide-semiconductor transistor (CMOS) device with a stressed channel.

Background technique

Current semiconductor technology is still making continuous efforts to reduce the volume of Very Large Scale Integration (VLSI) circuits. As the circuit size becomes smaller and the operation becomes faster, increasing the driving current of each element becomes an increasingly important issue. The device current is closely related to gate length, gate capacitance and carrier mobility. Shortening the polysilicon gate length, increasing gate capacitance, and accelerating carrier mobility can all increase device current. In terms of shortening the gate length, it means reducing the size of the components. In terms of increasing the gate capacitance, it can be achieved by reducing the thickness of the gate dielectric layer, increasing the gate dielectric constant, and so on. In order to improve the device current, how to increase the carrier mobility has also been extensively studied.

One of the existing methods to increase carrier mobility is to form a strained channel. The strain can increase the mobility of bulk electrons and holes. Strained channels can increase the performance of Metal Oxide Semiconductor (MOS) devices. The advantage of this technique is that it does not change the gate length, nor does it add additional circuit process or design.

When silicon is subjected to in-plane stress, the electron mobility at room temperature increases greatly. One method of applying stress is to use GeSi epitaxy with gradually changing Ge concentration as the substrate. A layer of silicon is formed on a relaxed silicon germanium layer, the layer of silicon is subjected to a stress, and then metal oxide semiconductor transistors are formed on the layer of silicon. Because the lattice constant of germanium silicide is larger than that of silicon, the silicon layer is subjected to one-dimensional tension, so that the mobility of carriers is also increased due to the stress.

In a device, stress can come from three different directions: a direction parallel to the channel length of the MOS transistor, a direction parallel to the channel width of the MOS transistor, and a direction perpendicular to the channel plane. The stress parallel to the length and width of the channel is called plane stress. According to research results, two-dimensional, planar tension can increase the performance of NMOS transistors, and compressive stress parallel to the channel length direction can increase the performance of PMOS transistors.

Contents of the invention

In view of this, the present invention mainly proposes a method to increase tension on the N-type MOS transistor and apply compressive stress to the P-type MOS transistor, so as to increase the performance of the MOS transistor.

The present invention provides a semiconductor element whose channel is stressed, and a manufacturing method thereof.

The present invention provides a semiconductor device, including a gate, a spacer, a buffer layer, and a source/drain region. The gate includes a gate electrode and a gate dielectric layer, and the gate dielectric layer is located under the gate electrode. The spacers form sidewalls of the gate electrode and the gate dielectric layer. The buffer layer is located on a semiconductor substrate, the buffer layer has a first part and a second part thicker than the first part under the gate dielectric layer and the spacer, wherein the upper surface of the first part of the buffer layer is thicker The upper surface of the second part located on the buffer layer is recessed and constitutes a recess. The source/drain regions are substantially aligned with the aforementioned spacers. The buffer layer has a lattice constant greater than that of the underlying substrate. The above-mentioned semiconductor device further includes a semiconductor capping layer located between the buffer layer and the gate dielectric layer of the second part, wherein the lattice constant of the semiconductor capping layer is smaller than that of the buffer layer.

In the semiconductor device of the present invention, the thickness of the buffer layer located in the second part is approximately between 2 nm and 50 nm.

According to the semiconductor device of the present invention, the semiconductor device is a P-type metal-oxide-semiconductor transistor, and the depth of the recess is less than about 50 nanometers.

According to the semiconductor element of the present invention, the semiconductor element is an N-type metal-oxide-semiconductor transistor, and the depth of the concave portion is approximately between 2 nanometers and 50 nanometers.

In the semiconductor device of the present invention, the buffer layer includes silicon, germanium, carbon or a mixture thereof.

In the semiconductor device of the present invention, the spacers extend below the gate dielectric layer, and the depth of the extension is less than about 30 nanometers.

In the semiconductor device of the present invention, the composition of the semiconductor capping layer includes silicon and germanium, and the germanium concentration of the semiconductor capping layer is lower than the germanium concentration of the buffer layer.

According to the semiconductor element of the present invention, the semiconductor element is a P-type metal oxide semiconductor transistor, and the thickness of the semiconductor capping layer is approximately between 0.5 nanometers and 20 nanometers.

In the semiconductor element of the present invention, the semiconductor element is an N-type metal oxide semiconductor transistor, and the thickness of the above-mentioned semiconductor covering layer is approximately between 0.6 nanometers and 25 nanometers, and the thickness of the above-mentioned semiconductor covering layer is greater than the above-mentioned The thickness of the semiconductor capping layer of the PMOS transistor.

In the semiconductor device according to the present invention, a part of the buffer layer is removed, so that the upper surface of the semiconductor substrate located under the first part of the buffer layer is recessed.

In the semiconductor element of the present invention, the thickness of the semiconductor covering layer of the P-type metal oxide semiconductor transistor is thinner than that of the N-type metal oxide semiconductor transistor.

Due to the difference in lattice constant, the semiconductor capping layer is subjected to a tension, and the buffer layer is subjected to a compression force. This mixed tension and compression structure makes the performance of the P-type metal-oxide-semiconductor transistor and the N-type metal-oxide-semiconductor transistor. be promoted.

The present invention provides a semiconductor element, which includes: a semiconductor substrate; a gate structure, including a gate electrode and a gate dielectric layer, and the gate electrode is located on the gate dielectric layer , the above-mentioned gate dielectric layer is located on the above-mentioned semiconductor substrate; a spacer is formed on the sidewall of the above-mentioned gate structure; and a buffer layer is located between the above-mentioned semiconductor substrate, the above-mentioned gate structure and the above-mentioned spacer , and the buffer layer has a first portion, a second portion thicker than the first portion, and a concave upper surface, which is substantially not covered by the gate structure, and constitutes a concave portion, substantially aligned with the outer edge of the spacer , and the lattice constant of the buffer layer is larger than the lattice constant of the semiconductor substrate.

In the semiconductor device according to the present invention, the lower surface of the concave portion extends downward to the semiconductor substrate.

The semiconductor device of the present invention further includes a semiconductor capping layer interposed between the buffer layer and the gate structure, and the lower surface of the recess extends downward to the semiconductor capping layer.

The present invention also proposes a method for forming a semiconductor device, including forming a buffer layer on a substrate, wherein the lattice constant of the buffer layer is different from the lattice constant of the substrate, and the buffer layer has a first portion The first part is thicker than the second part; forming a gate dielectric layer on the above-mentioned buffer layer; forming a gate electrode on the above-mentioned gate dielectric layer, patterning the above-mentioned gate dielectric layer and gate electrode respectively layer; forming a spacer on the sidewall of the gate to recess the buffer layer not located under the gate and the spacer; and forming source/drain regions substantially aligned with the spacer. The above-mentioned semiconductor element can be a P-type metal-oxide-semiconductor transistor with a recess depth not exceeding 50 nanometers; or an N-type metal-oxide-semiconductor transistor with a recess depth between 2 nanometers and 50 nanometers.

The lower surface of the recess can extend downwards to the semiconductor substrate, and in a preferred embodiment, the depth of the recess is no more than 30 nanometers below the gate dielectric layer.

The above method further includes forming a semiconductor capping layer, the lattice constant of the semiconductor capping layer is smaller than that of the buffer layer, and the semiconductor capping layer is located between the buffer layer and the gate dielectric layer. If the above-mentioned semiconductor device is a P-type metal oxide semiconductor transistor, the thickness of the semiconductor capping layer is approximately between 0.5 nanometers and 20 nanometers. If the semiconductor element is an NMOS transistor, the thickness of the semiconductor capping layer is about 0.6 nm to 25 nm. In a preferred embodiment, the thickness of the semiconductor capping layer of the PMOS transistor is smaller than the thickness of the semiconductor capping layer of the NMOS transistor.

The semiconductor element and the method for forming the semiconductor element in the present invention increase the tension on the N-type metal oxide semiconductor transistor, and apply compressive stress on the P-type metal oxide semiconductor transistor, so as to increase the performance of the metal oxide semiconductor transistor.

Description of drawings

1 to 5D are cross-sectional views showing the intermediate process of manufacturing a semiconductor device according to a preferred embodiment of the present invention;

6A to 6B are schematic channel diagrams of N-type and P-type metal oxide semiconductor transistors;

FIG. 7 is a graph showing the relationship between leakage current and drive current in an off-state.

Detailed ways

In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and understandable, a preferred embodiment is specifically cited below, and in conjunction with the accompanying drawings, the detailed description is as follows:

1 to 5D are cross-sectional views showing an intermediate process of manufacturing a semiconductor device according to a preferred embodiment of the present invention. In different drawings, the same numerals represent the same elements.

Figure 1 shows 2, 4 and 6 layers stacked in sequence of 3 layers. The base 2 is mostly a semiconductor material, and generally silicon is used as the base material. The substrate 2 can be a bulk silicon (bulk silicon) or an existing silicon-on-insulator (SOI) structure, wherein the silicon-on-insulator (SOI) is formed on a buried oxide later (BOX) an insulating layer.

The buffer layer 4 is mostly epitaxially grown on the substrate 2 . The lattice constant of the buffer layer 4 is probably greater than that of the substrate 2 . In a preferred embodiment, the composition of the buffer layer 4 includes silicon and germanium. Although the buffer layer does not only use silicon and germanium, the buffer layer 4 can also be written as a silicon germanium layer 4 for simplified representation. In other embodiments, the composition of the buffer layer 4 includes silicon, germanium and carbon. In a preferred embodiment, the thickness of the buffer layer is approximately between 2 nanometers (nm) and 50 nm.

A semiconductor capping layer 6 is epitaxially grown on the buffer layer 4 . In a preferred embodiment, the semiconductor capping layer 6 has a lattice constant smaller than that of the buffer layer 4 and is composed of silicon. In other embodiments, the composition of the semiconductor capping layer 6 may include silicon, germanium or other materials with similar properties, and the concentration of germanium is lower than the concentration of germanium in the buffer layer. Therefore, overall, the lattice constant of the semiconductor capping layer 6 is smaller than that of the buffer layer 4 . Although the semiconductor capping layer does not only use semiconductor materials, the semiconductor capping layer 6 can also be written as a silicon capping layer 6 for simplified representation. In a preferred embodiment, the thickness of the semiconductor capping layer 6 of the PMOS transistor is probably smaller than the thickness of the semiconductor capping layer 6 of the NMOS transistor. In a preferred embodiment, the thickness can be determined by the following two methods. (1) Simultaneously forming a semiconductor covering layer on the N-type and P-type metal oxide semiconductor transistors, and then reducing the thickness of the semiconductor covering layer of the P-type metal oxide semiconductor transistor. (2) Simultaneously forming a semiconductor covering layer with a first thickness on the N-type and P-type metal oxide semiconductor transistors, covering the P-type metal-oxide-semiconductor transistors (masking), and then on the N-type metal-oxide-semiconductor transistors A semiconductor capping layer with a second thickness is formed. The thickness of the semiconductor capping layer on the PMOS transistor is approximately between 0.5 nm and 20 nm. The thickness of the semiconductor capping layer on the NMOS transistor is about 0.6 nm to 25 nm.

Next, a gate dielectric layer 8 and a gate electrode 10 are formed, as shown in FIG. 2 . A gate dielectric layer is formed on the semiconductor capping layer 6, and the gate electrode is formed on the gate dielectric layer. Then pattern the desired position and size on the gate dielectric layer and the gate electrode layer, and then etch. In a preferred embodiment, the composition of the gate dielectric layer 8 includes silicon dioxide (SiO ₂ ), which can also be called the gate oxide layer 8 . In other embodiments, the composition of the gate dielectric layer 8 includes silicon oxynitride (oxynitride), silicon nitride (nitride), high-k material or other materials with similar properties. Although other materials such as metal, metal silicide or other materials with similar properties can be used for the gate electrode, in a preferred embodiment, the composition of the gate electrode 10 is mostly polysilicon.

FIG. 3 shows a pair of spacers 12 formed on the sidewalls of the gate dielectric layer 8 and the gate electrode 10 . The spacers 12 act as a self-aligned mask for source/drain electrodes in subsequent process steps. The source/drain will be described in detail later. The spacer 12 can be deposited by the existing method of blanket deposition (blanket deposit) a dielectric layer, and then anisotropic etching (anisotropic etch) from the lower surface of the dielectric layer to remove the above dielectric layer, what remains is the gap wall 12. It should be noted that when patterning the gate dielectric layer 8 and the gate electrode 10 , due to process variation, the spacers 12 may extend to the semiconductor capping layer 6 due to over-etching. Thus, the tensile strain of the semiconductor capping layer is reduced. Therefore, in a preferred embodiment, the overetched depth D _poly of the semiconductor capping layer 6 needs to be less than 30 nm.

The concave portion 9 is formed along the spacer wall 12, as shown in FIGS. 4A and 4B. FIG. 4A shows that, in a preferred embodiment, the lower surface of the recess 9 extends down to the semiconductor capping layer 6 and the buffer layer 4 and is located on the substrate 2 . The depth D _substrate of the recess 9 on the substrate 2 is less than about 50 nanometers. Free surfaces 5 , 7 are located on the side walls of the buffer layer 4 . Among them, the free surface means that there is almost no other material covering it during annealing (anneal), so that the material here can freely expand or shrink to its maximum limit, however, the final structure will still be covered with a layer Inter-dielectric layer (inter-layerdielectric). When the buffer layer 4 is annealed, the buffer layer 4 tends to relax and recover its lattice structure, so the buffer layer 4 expands. The free surfaces 5, 7 allow the buffer layer 4 to expand freely. In the subsequent annealing process, the buffer layer 4 will expand to the left of the free surface 5 and the right of the free surface 7 . Due to the balance of forces on both sides of the free surface, the buffer layer 4 generates compressive stress, while the semiconductor capping layer 6 generates tensile stress. In a preferred embodiment, the two free surfaces 5, 7 allow the buffer layer 4 to expand freely. In other embodiments, only one recess 9 is formed beside the gate dielectric layer 8 and the gate electrode, so there is only one free surface 5 or 7 . When the buffer layer is annealed, the buffer layer expands substantially in only one direction.

4A and 4B show that in a preferred embodiment, a recess 9 extends from spacers 12 on both sides to a shallow trench isolation (shallow trench isolation, STI) 11 . In other embodiments, as shown in FIG. 4C , the recess 9 has a width W which is smaller than the distance from the spacer 12 to the STI 11 . The width of W is probably so large that there is room for expansion when the buffer layer 4 is annealed.

Even if the sidewalls of the buffer layer 4 are not fully exposed, compressive stress can still be generated. In other embodiments, as shown in FIG. 4B , the semiconductor capping layer 6 is etched until part of the buffer layer 4 is recessed. The recess 9 is a depression in the buffer layer 4 with a depth D _layer . The D _layer of the PMOS transistor is approximately between 0 nm and 50 nm. The D _layer on the NMOS transistor is approximately between 2nm and 50nm.

Due to process variations, the recesses of the N-type MOS transistor and the P-type MOS transistor are not consistent. In some cases, the lower surface of the concave portion of the N-type metal-oxide-semiconductor transistor extends downward to the buffer layer 4, while the lower surface of the concave portion of the P-type metal-oxide-semiconductor transistor only extends downward to the semiconductor capping layer 6 and does not extend to buffer layer 4.

5A to 5D show the formation process of the source and drain electrodes 14 . In a preferred embodiment, the source and drain are formed by doping on the other side of the spacer 12 . In other embodiments, the source/drain is located in the concave portion of the semiconductor material not covered by the spacer 12 , and the semiconductor material is epitaxially grown in the concave portion according to the required impurity concentration.

According to the above-mentioned manufacturing method, several different embodiments can be derived. 5A and 5B show that the semiconductor capping layer 6 is omitted, so that the recess 9 is entirely located on the buffer zone 4, as shown in FIG. 5A; or the lower surface of the recess 9 extends downward to the substrate 2, as shown in FIG. 5B. In a preferred embodiment, due to the generation of compressive stress, FIG. 5A and FIG. 5B are more suitable for forming P-type metal oxide semiconductor transistors. 5C and 5D show that the semiconductor capping layer 6 is formed on the buffer layer 4 . Similar to FIGS. 5A and 5B , the recess 9 may be located at the same level as the semiconductor capping layer 6 and the buffer layer 4 , as shown in FIG. 5C ; or the lower surface of the recess 9 may extend downward to the substrate 2 , as shown in FIG. 5D . In a preferred embodiment, the recess 9 of the PMOS transistor is formed on the semiconductor capping layer 6 without extending to the buffer layer 4 .

6A and 6B respectively show the directions of the stresses 16 and 18 in the buffer layer 4 and the semiconductor capping layer 6 . Due to the small lattice constant of the substrate 2 , the buffer layer 4 is subjected to a compressive stress, as indicated by the direction of the arrow 16 . Due to the large lattice constant of the buffer layer 4 , the semiconductor capping layer 6 is subjected to a tensile stress, as indicated by the direction of the arrow 18 . FIG. 6A and FIG. 6B also show the channels 20 and 22 of the N-type and P-type metal-oxide-semiconductor transistors, respectively. Since the semiconductor capping layer 6 has a tensile stress, the performance of the NMOS transistor is improved. For PMOS transistors, the extra germanium lowers the valence band. For example, when the germanium concentration is about 10% to 30%, the valence band drops by about 100mV to 300mV. Therefore, the channel of the PMOS transistor tends to occur in the SiGe layer 4 rather than the Si capping layer 6 . Therefore, the channel region of the PMOS transistor has compressive stress, thereby increasing the performance of the PMOS transistor.

In PMOS transistors, silicon germanium (SiGe) is mostly used as the buffer layer. However, if the germanium silicide directly contacts the gate dielectric layer 8, a reliability problem will arise. Therefore, between the germanium silicide layer 4 and the gate dielectric layer 8 , a thin silicon capping layer 6 is generally used. If other gate dielectric materials are used, such as high dielectric constant materials, the thin silicon capping layer 6 can be omitted, just like the structure shown in FIG. 5A and FIG. 5B .

In the preferred embodiment, a semiconductor device contains mixed stress, ie, tensile stress in the semiconductor capping layer 6 and compressive stress in the buffer layer 4 . Due to the existence of mixed stress, the performance of N-type and P-type metal-oxide-semiconductor transistors can be improved. FIG. 7 shows the experimental data of the leakage current (leakage current) plotted against the operating current (drive current, or on current) of the PMOS transistor in the off-state (or off current). Line 26 is drawn for a device having a semiconductor capping layer and a buffer layer, and line 28 is drawn for a PMOS transistor formed on a silicon substrate. It is worth noting that, under the same off-current, the working current of the preferred embodiment proposed by the present invention is 15% higher than that of the PMOS transistor formed on the base silicon. There is also a 10% increase in NMOS transistors (not shown). Experimental data also shows that the improved I _on -I _off characteristics do not affect the performance of other characteristics.

Although the present invention has been described above through preferred embodiments, the preferred embodiments are not intended to limit the present invention. Those skilled in the art should be able to make various changes and supplements to the preferred embodiment without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention is subject to the scope of the claims. In addition, the scope of the present invention is not limited only to the process, equipment, manufacturing method, combination of things, means, methods or steps proposed in the embodiments. Those skilled in the art should understand that the processes, instruments, manufacturing methods, combinations of objects, means, methods or steps disclosed in the present invention, whether existing or to be developed, can all be based on the embodiments of the present invention Perform substantially the same function or achieve substantially the same result. Therefore, the scope of the claims includes process, apparatus, manufacturing method, combination of things, means, methods or steps.

The present invention proposes several advantageous method embodiments. Including: a method of forming a semiconductor device, including forming a buffer layer on a substrate, wherein the lattice constant of the buffer layer is different from the lattice constant of the substrate; forming a gate dielectric layer on the buffer layer above; form a gate electrode on the above-mentioned gate dielectric layer, pattern the above-mentioned gate dielectric layer and gate electrode layer; form a spacer on the side wall of the gate, so that it is not located And the buffer layer under the spacer is recessed; and the source/drain region is formed, roughly aligned with the spacer. The above-mentioned semiconductor element can be a P-type metal-oxide-semiconductor transistor with a recess depth not exceeding 50 nanometers; or an N-type metal-oxide-semiconductor transistor with a recess depth between 2 nanometers and 50 nanometers.

The lower surface of the recess can extend downwards to the semiconductor substrate, and in a preferred embodiment, the depth of the recess is no more than 30 nanometers below the gate dielectric layer.

The above method further includes forming a semiconductor capping layer, the lattice constant of the semiconductor capping layer is smaller than that of the buffer layer, and the semiconductor capping layer is located between the buffer layer and the gate dielectric layer. If the above-mentioned semiconductor device is a P-type metal oxide semiconductor transistor, the thickness of the semiconductor capping layer is approximately between 0.5 nanometers and 20 nanometers. If the semiconductor element is an NMOS transistor, the thickness of the semiconductor capping layer is about 0.6 nm to 25 nm. In a preferred embodiment, the thickness of the semiconductor capping layer of the PMOS transistor is smaller than the thickness of the semiconductor capping layer of the NMOS transistor.

A brief description of the symbols in the drawings is as follows:

2: base

4: buffer layer

5, 7: Free surface

6: Semiconductor cover layer

8: Gate dielectric layer

9: Concave

10: Gate electrode

11: Shallow trench isolation

12: spacer wall

14: Source/Drain

D _poly : Depth of overetching

W: Width of the recess

Claims (15)

1. A semiconductor element, characterized in that, the semiconductor element comprises: A gate, including a gate electrode and a gate dielectric layer, the gate dielectric layer is located under the gate electrode; a spacer formed on the sidewalls of the gate electrode and the gate dielectric layer; and A buffer layer located on a semiconductor substrate, the buffer layer has a first portion and a second portion thicker than the first portion under the gate dielectric layer and the spacer, wherein the buffer layer is located on the first portion of the buffer layer the surface is recessed relative to the upper surface of the second portion of the buffer layer and constitutes a recess; A source/drain region is substantially aligned with the spacer. 2. The semiconductor device according to claim 1, wherein a lattice constant of the buffer layer is larger than a lattice constant of the semiconductor substrate. 3 . The semiconductor device according to claim 1 , wherein the thickness of the buffer layer located at the second portion is between 2 nm and 50 nm. 4 . 4. The semiconductor device according to claim 1, wherein the semiconductor device is a P-type metal-oxide-semiconductor transistor, and the depth of the concave portion is less than 50 nanometers. 5 . The semiconductor device according to claim 1 , wherein the semiconductor device is an N-type metal-oxide-semiconductor transistor, and the depth of the concave portion is between 2 nanometers and 50 nanometers. 6. The semiconductor device according to claim 1, wherein the buffer layer comprises silicon, germanium, carbon or a mixture thereof. 7 . The semiconductor device according to claim 1 , wherein the spacer extends below the gate dielectric layer, and the extending depth is less than 30 nanometers. 8. The semiconductor device according to claim 1, further comprising a semiconductor capping layer interposed between the buffer layer of the second portion and the gate dielectric layer, and the crystal lattice of the semiconductor capping layer The constant is smaller than the lattice constant of the above-mentioned buffer layer. 9. The semiconductor device according to claim 8, wherein the composition of the semiconductor capping layer includes silicon and germanium, and the concentration of germanium in the semiconductor capping layer is smaller than that of the buffer layer. 10 . The semiconductor device according to claim 8 , wherein the semiconductor device is a P-type metal oxide semiconductor transistor, and the thickness of the semiconductor capping layer is between 0.5 nm and 20 nm. 11 . 11. The semiconductor element according to claim 10, wherein the semiconductor element is an N-type metal oxide semiconductor transistor, and the thickness of the semiconductor capping layer is between 0.6 nanometers and 25 nanometers, and the above-mentioned The thickness of the semiconductor capping layer is greater than the thickness of the semiconductor capping layer of the PMOS transistor described above. 12. A semiconductor element, characterized in that the semiconductor element comprises: a semiconductor substrate; A gate structure, including a gate electrode and a gate dielectric layer, and the gate electrode is located on the gate dielectric layer, and the gate dielectric layer is located on the semiconductor substrate; a spacer formed on the sidewall of the gate structure; and A buffer layer, the buffer layer is located between the semiconductor substrate, the gate structure and the spacer, and the buffer layer has a first part, a second part thicker than the first part, and a concave upper surface, substantially not It is covered by the gate structure and forms a concave portion, which is roughly aligned with the outer edge of the spacer, and the lattice constant of the buffer layer is larger than the lattice constant of the semiconductor substrate. 13 . The semiconductor device according to claim 12 , wherein a lower surface of the concave portion extends downward to the semiconductor substrate. 14 . 14. The semiconductor device according to claim 12, further comprising a semiconductor capping layer interposed between the buffer layer and the gate structure, and the lower surface of the recess extends downward to the semiconductor capping layer . 15. A method for forming a semiconductor element, characterized in that the method for forming a semiconductor element comprises: forming a buffer layer on a substrate, wherein the lattice constant of the buffer layer is different from the lattice constant of the substrate, and the buffer layer has a first portion and a second portion thicker than the first portion; forming a gate dielectric layer on the buffer layer; forming a gate electrode on the gate dielectric layer; patterning the gate dielectric layer and the gate electrode to form a gate; forming a spacer on the sidewall of the gate, so that a part of the buffer layer is under the gate and the spacer; forming a recess by recessing the buffer layer which is not located under the gate and the spacer; and A source/drain region is formed approximately aligned with the spacers.

Images